# Design a product recommendation system for my online store

A practical plan for personalized product recommendations: approach by data maturity, placements, cold-start handling, and how to measure lift.

- Topic: Data & Analysis
- Framework: Few-Shot
- Level: Standard
- Tags: E-commerce, Fill-in Template
- Source: https://www.thinkwithprompt.com/prompts/design-a-product-recommendation-system-for-my-online-store

## Prompt

```text
Help me design a personalized product recommendation system for my online store.

**Store and catalog:** [WHAT WE SELL, number of products, how often the catalog changes]
**Traffic and customers:** [monthly visitors, orders, share of logged-in or returning customers]
**Data available:** [browsing events, purchases, cart adds, ratings, product attributes, search queries]
**Platform and team:** [e.g. Shopify, custom stack; in-house data skills or none]
**Goal:** [higher average order value / conversion / repeat purchases / discovery]

Deliver:
1. **Approach by maturity:** start with what works for my data size, e.g. rules and bestsellers, "frequently bought together" from co-purchases, content-based similarity from product attributes, collaborative filtering, then learned ranking models. Say which stage fits me now and when to move up.
2. **Placements:** product page, cart, homepage, email, search, post-purchase, with the best recommendation type for each.
3. **Cold start:** how to handle new visitors and new products.
4. **Business rules:** stock and margin filters, diversity, avoiding recommending items just bought, promoted items, and fairness to smaller brands if relevant.
5. **Build vs. buy:** platform apps or vendors vs. a custom model, with trade-offs.
6. **Measurement:** offline metrics (precision, recall, coverage) and an A/B test plan with the business metric that decides success.
7. **Privacy:** consent for tracking, data retention, and clear explanations to customers.

If I want to build it myself, outline the data pipeline and give starter code in [Python/SQL].
```

## What to fill in

- e.g. Shopify, custom stack; in-house data skills or none
- Python/SQL

Free to copy and use. From ThinkWithPrompt (https://www.thinkwithprompt.com/), a library of 2,663 curated AI prompts.